JPH08170947A - Image defect discriminating processor - Google Patents

Abstract

PURPOSE: To perform the inspection due to full color printing with high accuracy and to inspect image defects of all of kinds by forming a plurality of comparison images on the basis of the reference image of an object free from an image defect obtained by an imaging means. CONSTITUTION: The reference image data taken by a CCD camera and image data to be inspected are stored in an image memory 402. The reference image data and image data to be inspected from an image input part 3 are supplied to an Lab (L*a*b* color system of CIE) image forming part 403 and a differential processing part 405 and an inspection range image forming part 404 forms inspection range data from the reference image data stored in the memory 402. An image comparative operation part 408 performs comparison processing using the reference image data and image data to be inspected stored in the memory 402. The inspection range image data obtained in the image forming part 404 is supplied to a defective pixel extraction part 410 and an inspection discriminating part 411 discriminates whether the printing state of a CD (compact disk) to be inspected is normal on the basis of the number of defective pixels and the number of defects.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image which is automatically inspected by using image processing technology for defects such as stains and scratches on prints generated in the production process of printed matter and stains and unevenness on the surface in manufacturing industrial products. The present invention relates to a defect discrimination processing device, and particularly to an image defect discrimination processing device suitable for inspecting a color printed matter.

[0002]

2. Description of the Related Art For example, a compact disc (hereinafter referred to as C
Say D. Since the surface of () is usually mirror-like due to aluminum vapor deposition and surface coating, it is totally reflected and a thin film thickness causes moiré with respect to illumination. C like this
A signal is written on the surface of D at the time of production of a CD, and a label such as a manufacturer name and a title is printed at a predetermined position. Then, the characters, patterns, etc. printed on the surface of the CD are inspected for deviations, chips, stains, stains, pock marks, and the like.

Generally, in an image defect discrimination processing apparatus for inspecting a state printed on the surface of a CD or the like, an image processing technique is used, and a reference image stored in a memory in advance and an object to be inspected are inspected. An image defect is discriminated by an image comparison method for comparing with an image, and a print state is inspected.

However, when performing such an inspection, the CD
The surface was photographed by a television (hereinafter referred to as TV) camera having a rectangular screen. Therefore, the CD
In the case of a disk-shaped object such as a disk whose central portion is open, the four corners and the central portion of the screen become ineffective, and the effective use area of the screen is reduced. As a result, there is a problem that one pixel becomes large and the image quality becomes rough, and thus the resolution is lowered. Further, when an image is captured by a monochrome camera, for example, colors such as yellow, sky blue, and pink have almost the same gray level in monochrome, and there is a problem that multicolor printing cannot be inspected.

Therefore, as disclosed in Japanese Patent Application No. 2-332089, for example, an image defect discrimination processing device using a line sensor camera has been proposed.

That is, the image defect discrimination processing device photographs, for example, a rotating body that fixes a circular body such as a CD and continuously rotates by one rotation, and a portion above the circular body along a radius line of the circular body. And a detection unit having a line sensor camera provided to do so, and an image processing unit that determines the quality of printing on the surface of the circular body by comparing and collating the captured image of the circular body with a reference previously captured by the same line sensor camera. And a monitor unit for displaying images of a circular body and a reference, and the line sensor camera is a color line sensor camera.

As described above, by rotating the circular body and photographing the radial line portion with the line sensor camera, it is possible to obtain a high-resolution image quality and perform an accurate inspection. Further, by using a color line sensor camera as the line sensor camera, it is possible to accurately perform color inspection in color printing.

However, when the line sensor camera as described above is used, it is necessary to provide a rotating body or the like for rotating a circular body, and it becomes a large scale to directly incorporate the image defect discrimination processing apparatus into the printing apparatus. There is also a problem that it is not possible to cope with a high-speed printing cycle in the printing apparatus.

On the other hand, as disclosed in Japanese Patent Application No. 3-264442, an image defect discrimination processing device has been proposed for the purpose of making it possible to detect the type of image defect. That is, for example, the image defect determination processing device predetermines a difference degree image data storage unit that stores difference degree data for each pixel between a reference image and an image to be inspected in correspondence with an address, and a stored difference degree data corresponding address. The small area address generation unit that divides the divided small area and the difference data and the first threshold value for each small area are compared, and the number of difference data that exceeds the first threshold value is set in advance. And a dissimilarity pixel calculation determination unit that determines dissimilarity data that exceeds the first threshold value of the small area as a pixel defect when the second threshold value set in step 1 is exceeded.

[0010]

However, the image defect discrimination processing device as described above cannot inspect all kinds of image defects, and in particular, inspects pockmarks, which are characteristic defects of CDs. I couldn't.

Therefore, the present invention has been made in view of the above-mentioned conventional circumstances, and has the following objects.

That is, an object of the present invention is to provide an image defect discrimination processing apparatus for inspecting all kinds of image defects by performing inspection by full color printing with high accuracy.

[0013]

In order to solve the above-mentioned problems, an image defect discrimination processing device according to the present invention is based on a reference image of an object having no image defect obtained by an image pickup means for capturing a color image. The comparison image generating means for generating a plurality of comparison images, the image extracting means for extracting the image of the inspection range from the inspected image to be discriminated and the reference image obtained by the image capturing means, and the image extracting means An information image generating unit that generates an information image in which all the pixels of the reference image in the inspection range are replaced with information corresponding to a predetermined characteristic for each pixel, and a plurality of comparison images generated by the comparison image generating unit. A determination unit that determines a comparison image based on the information image generated by the information image generation unit, a comparison image determined by the determination unit, and an inspection obtained by the image extraction unit. A discriminating unit for discriminating an image defect of the image to be inspected by comparing the image to be inspected in a range to obtain the degree of difference for each pixel, and comparing the degree of dissimilarity with an arbitrarily set threshold value. It is characterized by

[0014]

In the image defect discrimination processing apparatus according to the present invention, the comparison image generating means generates a plurality of comparison images based on the reference image having no image defect obtained by the image capturing means for capturing the color image. The image extracting means extracts an image of the inspection range from the inspected image to be discriminated and the reference image obtained by the imaging means. The information image generating means generates an information image in which all the pixels of the reference image in the inspection range obtained by the image extracting means are replaced with information corresponding to a predetermined characteristic for each pixel. The determining means determines a comparative image from the plurality of comparative images generated by the comparative image generating means based on the information image generated by the information image generating means. The discriminating means compares the comparison image determined by the determining means with the inspected image in the inspection range obtained by the image extracting means to obtain the degree of difference for each pixel, and the degree of difference is arbitrarily set. The image defect of the image to be inspected is discriminated by comparing with the threshold value.

[0015]

An embodiment of the present invention will be described below with reference to the drawings.

The image defect discrimination processing apparatus according to the present invention is an apparatus for inspecting the color printing state of the surface of a compact disc (hereinafter referred to as a CD) by a fully automatic inspection system, for example, as shown in FIG. , CD mount 11 and C
An image pickup section 12 for taking in a color image of the surface of a reference image creation CD or an inspected CD installed on the D mounting table 11, and an image input section 3 to which color image data obtained by the image pickup section 12 is supplied. The image processing unit 4 is provided with the RGB reference image data or the RGB inspected image data obtained by the image input unit 3, and the image output unit 5 to which the inspection result is supplied from the image processing unit 4.

The CD mounting table 11 and the image pickup section 12 are installed in the lighting device 1. The illumination device 1 is, for example, a dedicated illumination device in which other noise-like images are not reflected by the aluminum vapor deposition surface of the CD, and although not shown, it is perpendicular to the oblique illumination suitable for the inspection of the color printing surface. It has an epi-illumination. A CD (not shown) is mounted on the CD mounting base 11.
A fixing tray is mounted, and one image pickup unit 12 is installed directly above the CD fixing tray.

The image pickup section 12 includes a camera lens 121 and a C lens.
CD (Charge coupled device)
It is composed of a color camera 122. Here, since the image resolution of the single-plate CCD color camera is reduced by half, a three-plate CCD color camera is used as the CCD color camera 122 in this embodiment. Therefore, CCD
The color camera 122 is an RGB (red-green-blue) separation color camera and can inspect the color of color printing by processing an image by dividing it into three colors of RGB.
A camera lens 121 is attached below the CCD color camera 122. Here, the focal length of the camera lens 121 is 25 mm.

The CD mounting table 11 has a C having no image defect.
D, or a CD to be inspected is installed, and during inspection, the installed CDs are sequentially and automatically conveyed directly below the CCD color camera 122 by rotating the CD mounting table 11. ing. Also, after the inspection is completed,
It is designed to be sent out automatically.

As shown in FIG. 2, the image processing unit 4 includes an image memory 402, a Lab image generating unit 403, a differential processing unit 405, an inspection range image generating unit 404, a characteristic flag image generating unit 406, and A correction processing unit 407, an image comparison calculation unit 408, a chromaticity matching degree calculation unit 409, a defective pixel extraction unit 410, and an inspection determination unit 411 are provided. The RGB output from the image input unit 3 is stored in the image memory 402.
Reference image data and RGB inspected image data are stored. The image memory 402 is, for example, a main CPU (centr, not shown) that controls the entire system.
DSP (digital signal processor) that functions as an al processing unit, and large-scale integrated circuit (LSI: Large Scale) for image processing
e Integration). Also, 512 pixels x 512 pixels x 8
It is a memory having a capacity to store 28-bit image data for 28 screens. Therefore, a large amount of reference image data can be permanently installed in the image memory 402. This enables high-speed and high-precision inspection. Table 1 shows the allocation of the image memory of 28 screens.

[0021]

[Table 1]

Lab image generation section 403 and differentiation processing section 40
The RGB reference image data and the RGB inspected image data output from the image input unit 3 are supplied to 5. The inspection range image generation unit 404 generates inspection range image data from the RGB reference image data stored in the image memory 402. The characteristic flag image generation unit 406 includes a differentiation processing unit 40.
The edge emphasis reference image 1 data or the edge emphasis reference image 1 data obtained in 5 is supplied.

The inspection range image data obtained by the inspection range image generation section 404 is supplied to the correction processing section 407.
Further, the correction processing unit 407 performs positional deviation correction based on the RGB reference image data and the RGB inspected image data stored in the image memory 402.

The image comparison / calculation unit 408 has inspection range image data obtained by the inspection range image generation unit 404, edge emphasis reference image data and edge emphasis inspected image data obtained by the differentiation processing unit 405, and characteristics. Flag image generator 4
The characteristic flag image data obtained in 06 is supplied. Further, the image comparison calculation unit 408, in accordance with the characteristic flag image data, the RGB reference image data and the RGB inspection image data stored in the image memory 402, or the edge emphasis reference image data and the edge emphasis inspection image data. To perform the comparison process.

The chromaticity matching degree calculation unit 409 has a Lab reference image data and a Lab inspected image data, which will be described later, obtained by the Lab image generation unit 403, and an inspection range image generation unit 404.
The inspection range image data obtained in step S4 and the comparison result in the image comparison calculation unit 408 are supplied.

To the defective pixel extracting section 410, the inspection range image data obtained by the inspection range image generating section 404 and the calculation result of the chromaticity matching degree calculating section 409 are respectively supplied. The defective pixel information obtained by the defective pixel extraction unit 410 is supplied to the inspection determination unit 411.

Here, FIG. 3 is a processing block diagram showing the inspection processing in the image defect discrimination processing apparatus shown in FIG. 1 and FIG. The reference numerals given to the processing blocks shown in FIG. 3 correspond to the reference numerals given to the constituent elements shown in FIGS. 1 and 2. Hereinafter, description will be made with reference to FIGS.

The reference image generation process S1 will be described below.

The reference image generated from the CD having no image defect is a grayscale image in which the RGB output of the CCD color camera 122 is digitized as it is, a differential image in which the edge portion is differentially emphasized, and the RGB image is chromaticity by a look-up table. It is composed of a Lab (CIE L ^* a ^* b ^* color system) image converted into coordinates.

Specifically, the reference image is the R of the camera.
Signal digital image (512 pixels x 512 pixels x 8 bits), camera G signal digital image (512 pixels x 5 bits)
12 pixels x 8 bits), camera B signal digital image (512 pixels x 512 pixels x 8 bits), CIE L ^*
a ^* b ^* color system L ^* digital image (512 pixels × 512
Pixel x 8 bits), CIE L ^* a ^* b ^* color system a ^* digital image (512 pixels x 512 pixels x 8 bits), CI
L ^* a ^* b ^* color system b ^* digital image of E (512 pixels x
512 pixels x 8 bits, inspection range image (512 pixels x
512 pixels × 1 bit), characteristic flag image (512 pixels × 512 pixels × 3 bits), edge enhanced image 1 (512
Pixel x 512 pixels x 8 bits), edge enhanced image 2 (5
12 pixels × 512 pixels × 8 bits), and the pixel address of the image, that is, the coordinate, is at the same position on all screens.

When generating the reference image as described above, C
A CD for creating a reference image is placed on the D mounting table 11. Then, the reference image creating CD, that is, the CD having no image defect is installed directly under the CCD color camera 122. A color image printed on the surface of the CD is captured by the CCD color camera 122 by the camera lens 121. The color image obtained by the CCD color camera 122 is supplied to the image input unit 3. At this time, the aperture of the camera lens 121 when photographing the reference image generating CD is slightly opened unlike when photographing the inspected CD. This causes the image to be digitized to be slightly saturated.

The image input section 3 includes a CCD color camera 12
The color image obtained in 2 is converted into an R signal digital image, a G signal digital image, and a camera B signal digital image (hereinafter referred to as RGB reference image), and RGB reference image data is converted. The image memory 4 is supplied to the Lab image generation unit 403 and the differential processing unit 405, respectively.
Output to 02. The reference image memory 402a of the image memory 402 stores the RGB reference image data output from the image input unit 3.

The Lab image generation unit 403 is composed of a look-up table (LUT) 403a for L ^* a ^* b ^* and an arithmetic processing unit 403b, and the RGB reference image data from the image input unit 3 is transferred to the CIE. L ^* a ^* b ^* color system. The conversion color used when converting to this L ^* a ^* b ^* color system is not the CIE conversion formula generally used,
Use the NTSC conversion formula.

That is, the RGB reference image data is color-measured by the arithmetic expression of X = 0.6067R + 0.1736G + 0.2001B Y = 0.988R + 0.5868G + 0.1144B Z = 0.0661G + 1.1150B.

[0035] Thus, X, Y, a value obtained by measuring with Z, with perfect reflecting diffuser tristimulus values Xo, Yo, and ^{Zo, L * = 116 (Y} / Yo) 1/3 -16 a * = 500 ((X / Xo) ¹ / 3- (Y / Yo) ^1/3 ) b ^* = 200 ((X / Xo) ¹ / 3- (Z / Zo) ^1/3 ) L ^* a ^* b ^* color system, that is, L ^*
It is converted into a digital image, an a ^* digital image, and a b ^* digital image (hereinafter referred to as Lab reference image). Here, for example, when Y / Yo ≦ 0.008856, L ^* = 903.25 (Y / Yo) ^1/3 .

The calculation as described above is performed by the calculation processing unit 403b.
Done in. In addition, the value of each conversion formula is set in advance in the LUT 402a.
And a Lab image is created in real time when the reference image is created. And the Lab generated in this way
The image data is supplied to the chromaticity image coincidence calculation unit 409.

The inspection range image generator 304 extracts an image of the inspection range from the RGB reference image data stored in the image memory 402. Here, the image capturing unit 12 captures an image of the CD mounted on the CD mounting base 11, as described above.
That is, the RGB reference image data stored in the image memory 402 is an image of the CD and the CD mounting table 11 together, and it is not necessary to inspect other than the CD portion. Therefore, since the contrast between the CD and the CD mounting table 11 is different, the CD portion is extracted from the RGB reference image data and the center of gravity position thereof is calculated. Then, the color-printed concentric circle range is obtained from the calculated barycentric position. The concentric range thus obtained is supplied to the correction processing section 407, the image comparison calculation section 408, the chromaticity matching degree calculation section 409, and the defective pixel extraction section 410 as inspection range image data. The differential processing unit 405
It has a product-sum calculator (not shown). This product-sum calculator performs differential emphasizing processing to extract the edge portion of the RGB image data from the image input unit 3, and the differential image 1
(Hereinafter, referred to as edge enhancement reference image 1) is generated, and the edge enhancement reference image 1 is supplied to the image comparison calculation unit 408 and the characteristic flag image generation unit 406, respectively. The edge emphasis reference image 1 is used as a congestion degree of a printing state in which it is determined that a portion where thin lines are densely printed has a high degree of density and a portion where a large range of the same color ink is printed is low. To be done.

Here, for example, when fine lines are densely printed on the aluminum vapor deposition surface of a CD, for example, small fine characters such as a title may be printed. In the color inspection of the portion where the lines are dense as described above, an inspection error is more likely to occur than in the case where the lines are not dense.
Therefore, in such a case, the differentiation processing unit 405 generates the edge emphasis reference image 2 in which the threshold value is changed with respect to the edge emphasis reference image 1, and the edge emphasis reference image 2 is used as the edge emphasis reference image 1. As a characteristic flag image generation unit 406
Supply to. As described above, since the inspection with a small error is performed using the density, the inspection accuracy can be improved.

The characteristic flag image generation unit 406 classifies the edge-enhanced reference image data 1 obtained by the differential processing unit 405 into three types of characteristics for each pixel.

More specifically, first, the CD surface is
Usually, aluminum vapor deposition and surface coating make it mirror-like and totally reflected, and due to its thin film thickness, moiré occurs for illumination. A portion where such moire occurs is an aluminum vapor deposition surface, that is, a portion where color printing is not performed. Here, when a blur of printing ink or the like occurs on the aluminum vapor deposition surface, a defect is easily detected by the difference between the edge-enhanced image data obtained from the reference image creation CD and the inspected CD. Therefore, the inspection process for the aluminum-deposited portion may be simplified. Therefore, a pixel obtained by photographing the portion where aluminum is vapor-deposited is set as the first characteristic.

Further, as in the case where moire is generated on the aluminum vapor deposition surface, a high-order frequency is generated at the boundary portion due to the spatial periodic pattern, that is, the pattern pitch of the CCD pixels, so the image at the boundary portion is , No clear color appears. That is, the color printing cannot be inspected at the boundary.
Therefore, a pixel obtained by photographing the boundary between the portion vapor-deposited with aluminum and the portion printed with color ink is used as the second characteristic.

Further, a pixel obtained by photographing a portion printed with color ink is set as a third characteristic, and these three types of characteristics are indicated by flag information.

Therefore, all the pixels of the edge emphasis reference image are classified into corresponding characteristics for each pixel and replaced with flag information. The characteristic flag image data generated in this way is supplied to the image comparison calculation unit 408.

As described above, the digital image data (RG) of R, G, B signals for the reference image generation CD is recorded.
B reference image data), CIE L ^* a ^* b ^* color system L ^* ,
a ^* , b ^* digital image data (Lab reference image data)
Then, the inspection range image data, the characteristic flag image data, and the edge emphasis reference image data 1 are generated.

Next, the inspection image generation processing S2 will be described.

The inspected image generated from the inspected CD is a digital image of the R signal of the camera (512 pixels × 512 pixels × 8 bits) and a digital image of the G signal of the camera (512 pixels).
Pixel x 512 pixels x 8 bits), digital image of camera B signal (512 pixels x 512 pixels x 8 bits), CI
E L ^* a ^* b ^* color system L ^* digital image (512 pixels x
512 pixels × 8 bits), CIE L ^* a ^* b ^* color system a ^* digital image (512 pixels × 512 pixels × 8 bits), CIE L ^* a ^* b ^* color system b ^* digital image (5
12 pixels x 512 pixels x 8 bits), edge enhanced image 1
(512 pixels x 512 pixels x 8 bits),
The pixel address of the image, that is, the coordinate is the same position on all screens.

When the inspection image as described above is generated,
The CD to be inspected is placed on the CD mounting table 11. The CD mounting base 11 rotates, and the inspection target C mounted on the CD mounting base 11 is rotated.
D is sequentially and automatically conveyed right below the CCD color camera 122.

As in the case of the reference image creating process 1, CC
The color image obtained by the D color camera 122 is supplied to the image input unit 3.

The image input section 3 includes a CCD color camera 12
The color image obtained in 2 is converted into a digital image of the R signal, a digital image of the G signal, and a digital image of the B signal of the camera (hereinafter, referred to as RGB inspected image), and the RGB inspected image is converted. The data is converted into the Lab image generation unit 40.
3 and the differential processing unit 405, and outputs them to the image memory 402. The inspected image memory 402b of the image memory 402 includes the RGB output from the image input unit 3.
The inspection image data is stored.

The Lab image generating unit 403 performs the same processing as in the case of the reference image creating process 1 with RGB from the image input unit 3.
The inspected image data is converted into L ^{* d} , a ^{* d} , b ^{* d} digital image data (hereinafter referred to as Lab inspected image data), and the Lab image data is supplied to the chromaticity image coincidence calculation unit 409. .

The differential processing section 405 uses the reference image creation processing 1
In the same manner as in the above case, differential processing is performed to extract the edge portion of the RGB inspected image data from the image input unit 3 to generate the edge enhanced inspected image 1, and the edge enhanced inspected image 1 is subjected to image comparison calculation. Supply to the unit 408.

As described above, digital image data of R, G, and B signals (RGB inspected image data) and L ^* a ^* b ^* color system L ^{* d} , a of the CIE for the inspected CD.
^{* d} , b ^{* d} digital image data (Lab inspected image data) and edge-enhanced inspected image 1 data are generated.

Next, the process of inspecting the printing state with the reference image and the image to be inspected obtained as described above will be described.

First, for a CD automatically conveyed by a machine, the vertical position is the position under the CCD color camera 122 at the time of inspection.
Both the horizontal axis and the rotation direction are misaligned. When the reference image and the image to be inspected are compared in such a state that the positional deviation is included,
There arises a problem that a difference in images caused by the positional deviation is erroneously determined as a defect. Therefore, the correction processing unit 4
In 07, when the inspected CD is conveyed under the CCD color camera 122 and the RGB inspected image data of the inspected CD is taken into the image memory 402, the mutual positional deviation amount is calculated and corrected.

That is, the correction processing unit 407 is supplied with the inspection range image data from the inspection range image generation unit 404.
The correction processing unit 407 stores the R stored in the image memory 402.
The GB reference image data and the RGB inspected image data are normalized and cross-correlated to detect a position having the highest degree of coincidence within the inspection range indicated by the inspection range image data. Its detection location and R
The amount of positional deviation between them is calculated by comparing with the GB reference image data. The calculated positional deviation amount is supplied to, for example, an image movement processing unit (not shown), and the image movement processing unit
The image is moved based on the amount of positional deviation from the correction processing unit 407.

The image comparison / calculation unit 408 compares the mutual pixels of the reference image and the inspection image within the inspection range indicated by the inspection range image data from the inspection range image generation unit 404,
All the pixels are classified into pixels that are determined to be normal, pixels that are suspicious, and pixels that are determined to be defective.

First, based on the characteristic flag image data from the characteristic flag image generation unit 406, it is determined which image data of the reference image is to be used for comparison based on the flag information distributed to each pixel.

For example, when the flag information of the target pixel indicates that the pixel where the aluminum vapor-deposited portion is photographed is taken, the edge emphasis reference image 1 data and the edge emphasis inspected image 1 data from the differential processing section 405. To use. This edge emphasis reference image 1 data and edge emphasis inspected image 1
When comparing the data with the data, the comparison gain is changed according to the density of the image data, and if the edge emphasis reference image 1 data M and the edge emphasis inspected image 1 data T are used, the comparison arithmetic expression is Z = | M− T | − (MT) / k−γ. In this comparison arithmetic expression, the first term is the edge emphasis reference image 1 data M and the edge emphasis inspected image 1 data T.
, And the second and third terms represent the function that changes the gain. Whether or not each pixel is normal is determined from the correlation degree between the comparison result Z obtained by this comparison arithmetic expression and the edge emphasis reference image 1 data M.

When the flag information of the target pixel indicates that the pixel printed by the color ink is a pixel, the RGB reference image data and the RGB inspected image data stored in the image memory 402 are used. . Then, the difference between the RGB reference image data and the RGB inspected image data is calculated for each pixel, and it is determined whether the pixel is a normal pixel or not according to the level of the difference. Here, the boundary between colors is
All pixels are judged to be suspicious. In addition, it is possible to deal with uneven printing and printing in which the gradation is gradually reduced.
Unless the degree of coincidence is considerably high, it is not judged as a normal pixel.

If the flag information of the target pixel indicates that the pixel is a picture of the boundary between the aluminum vapor-deposited portion and the portion printed with color ink, the edge emphasis reference image 1
Data and edge-enhanced inspected image 1 data are used. The comparison arithmetic expression in this case is similar to the case where it is shown that the pixel is a photographed portion of the aluminum vapor deposition described above, but the normality of each pixel used in making a comprehensive judgment is The weighting coefficient is set to be lower than that in the case where it is shown that the pixel is an image of a portion where aluminum is vapor-deposited. This reduces the influence of the inspection of the boundary portion.

As described above, the comparison is performed for each pixel using the image data based on the flag information, and the result of determining whether the pixel is a normal pixel or not is supplied to the chromaticity coincidence calculation unit 409. .

The chromaticity matching degree calculation unit 409 determines that the target pixel is a color within the inspection range indicated by the inspection range image data from the inspection range image generation unit 404 according to the comparison result from the image comparison operation unit 408. The color information is compared with respect to all the pixels which are the pixels obtained by photographing the portion printed with the ink and whose comparison result is determined to be the suspicious pixel.

That is, La from the Lab image generator 403
b With the reference image data (= L ^* , a ^* , b ^* ) and the Lab inspection image data (= L ^{* d} , a ^{* d} , b ^{* d} ), the spatial distance difference ΔE between the reference image and the inspection image. ΔEab ^* = ((ΔL ^* ) ² + (Δa ^* ) ² + (Δb ^* ) ² ) ^1/2 ΔL ^* = L ^{* d−} L ^* Δa ^* = a ^{* d−} a ^* Δb ^* = It is obtained by the arithmetic expression b ^{* d−} b ^* .

Further, in order to carry out an inspection in consideration of color bleeding and the like, the color difference in the peripheral 8 connection direction of the target pixel is multiplied by each arbitrary coefficient, and the total value thereof is used as the color difference information of the target pixel. The target pixel is classified into a pixel determined to be normal and a pixel determined to be defective according to the magnitude of the color difference information thus obtained.

As described above, it is determined whether or not many suspicious pixels are present also in the peripheral pixels of the target pixel which is determined to be the suspicious pixel, and when many suspicious pixels are present, it is determined as a defective pixel. To do. The result of such comparison is supplied to the defective pixel extraction unit 410.

The defective pixel extraction unit 410 determines that there is a defect within the inspection range indicated by the inspection range image data from the inspection range image generation unit 404 based on the comparison result from the chromaticity matching degree calculation unit 409. Only the determined pixels are replaced with the high level of the binarized image. Then, the image is labeled to classify the defective image into small regions, and the number of defective pixels, that is, the defect area and the defect coefficient are obtained for each labeled small region.

Here, the image data is classified into three types of characteristics by the flag information, and different coefficients are given to the pixels judged to be defective. The coefficients are added in the small area, and the added result is taken as the defect coefficient. The number of defective pixels and the defect coefficient thus obtained are supplied to the inspection determination unit 411.

The inspection discriminating section 411 has a defective pixel extracting section 41.
Based on the number of defective pixels and the defect coefficient from 0, it is determined whether the print state of the inspected CD is normal. Then, the result of this determination is supplied to the image output unit 5 shown in FIG.

As described above, in the present embodiment, the image data to be compared is determined according to the characteristics of each pixel, and the printing state is inspected using the determined image data. It is possible to inspect all kinds of defects such as blemishes, fading and pock marks. The image defect discrimination processing device described above can be applied to, for example, a device for automatically inspecting a defect of a mask pattern in the manufacture of an LSI by using an image processing technique.

[0070]

In the image defect discrimination processing apparatus according to the present invention, the comparison image generating means generates a plurality of comparison images based on the reference image of the object having no image defect obtained by the image capturing means for capturing the color image. . The image extracting means extracts an image of the inspection range from the inspected image to be discriminated and the reference image obtained by the imaging means. The information image generating means generates an information image in which all the pixels of the reference image in the inspection range obtained by the image extracting means are replaced with information corresponding to a predetermined characteristic for each pixel. The determining means determines a comparative image from the plurality of comparative images generated by the comparative image generating means based on the information image generated by the information image generating means. The determining means compares the comparison image determined by the determining means with the inspected image of the inspection range obtained by the image extracting means to obtain the degree of difference for each pixel,
The image defect of the image to be inspected is discriminated by comparing the above-mentioned difference degree with a threshold value set arbitrarily. As a result, the inspection by full-color printing can be performed with high precision, and it is possible to inspect all kinds of image defects such as print stains, stains, blurs, and pock marks.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image defect discrimination processing device according to the present invention.

FIG. 2 is a block diagram showing a configuration of an image processing unit of the image defect determination processing device.

FIG. 3 is a processing block diagram for explaining processing of the image defect determination processing device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Lighting device 3 Image input unit 4 Image processing unit 5 Image output unit 11 CD mounting base 12 Imaging unit 402 Image memory 403 Lab image generation unit 404 Inspection range image generation unit 405 Differentiation processing unit 406 Characteristic flag image generation unit 407 Correction processing unit 408 Image comparison calculation unit 409 Chromaticity coincidence calculation unit 410 Defective pixel extraction unit 411 Inspection determination unit

Claims (1)

[Claims] 1. A comparison image generating means for generating a plurality of comparison images based on a reference image of a target having no image defect obtained by the image capturing means for capturing a color image, and an object of discrimination target obtained by the image capturing means. Image extracting means for extracting an image in the inspection range from the inspection image and the reference image, and all pixels of the reference image in the inspection range obtained by the image extracting means are replaced with information corresponding to a predetermined characteristic for each pixel. An information image generating means for generating the information image, a determining means for determining a comparative image based on the information image generated by the information image generating means from the plurality of comparative images generated by the comparative image generating means, The degree of difference for each pixel is obtained by comparing the comparison image determined by the determining means with the image to be inspected in the inspection range obtained by the image extracting means, and the degree of difference is arbitrarily set. Image defect determination processing device characterized by comprising a discrimination means for discriminating an image defect inspection image by comparing the threshold was.